A comparison of binaural recordings performed in six different room configurations

Sammanfattning: Children in Swedish preschools are exposed to high noise levels that can cause hearing damage or voice related problems among both teachers and students, this since they have to raise their voice in order to make themselves heard in a noisy environment. Because of this it is of interest to be able to predict the sound environment in preschools. A way to do this is through convolving binaural recordings of typical preschool noise with simulated binaural impulse responses of a preschool’s premises. In order for this to work the binaural recordings has to be performed in an anechoic chamber. However, it is not always considered safe to perform binaural recordings of preschool children in an anechoic environment since the floor often consists of a net. A semi-anechoic environment, however, has a real floor and could possibly be created on a preschools own premises by applying sound absorbing elements on the walls. But will binaural recordings performed in a semi-anechoic environment give the same results as binaural recordings performed in an anechoic environment after convolution? It is assumed that if no difference is perceived between the binaural recordings performed in the two different environments there should not be a perceived difference between them after convolution as well. In order to investigate this, binaural recordings were performed in six different room configurations with different absorption, from fully reverberant to anechoic. The recordings were performed by using an artificial head. Room acoustic parameters, such as reverberation time, speech clarity and sound pressure level was measured in all room configurations at the same places as the artificial head and the loudspeaker had been placed. The results of the measurements showed that the reverberation time was the shortest in room configuration C, which represented a semi-anechoic chamber. In this room configuration there was no values for the speech clarity for frequencies above 1000 Hz, this suggests that there is a lack of late reflections. A listening test was then performed, all room configurations were compared to the anechoic recording in order to investigate if there was a perceived difference and how large that difference was in that case. The results of the listening test were then analyzed statistically through the method of t-test. Results showed that there was a significant difference between binaural recordings performed in all of the six room configurations and binaural recordings performed in an anechoic chamber. To characterize the absorption in each room an average absorption coefficient was calculated with the help of Eyring’s formula. The calculations showed that the value of the average absorption coefficient was largest for room configuration C where it had a value of approximately 0,7. In anechoic chambers the average absorption coefficient usually has a value around 1. The significant difference between how binaural recordings performed in room configuration C and binaural recordings performed in an anechoic chamber are perceived suggests that it is 4 the early reflections that are of importance for the perceived difference. This because there is a lack of late reflections in this room configuration for frequencies above 1000 Hz.